feat: integrate chart storage optimization into orderbook v2.7.218

[jackhai9]

The native Binance TradingView mirror opens a destination transaction for every stored record, including unchanged records. Orderbook v2.7.218 now includes a narrowly scoped writer that uses one real IndexedDB transaction per admitted mirror batch and skips equal records. Users keep the existing orderbook installer; there is no additional installation entry.

The hook starts in the top-level page at document-start and requires an exact match of the pinned native module before its first execution. Unsupported input and pre-transaction rejection preserve the original expression. A transaction failure aborts the batch, retires optimization and never replays uncertain writes. Ordinary saves, Basic settings, source reads, clear and financial operations retain their native paths. Stop drains accepted work. The complete orderbook also waits for creation of the HTML root before its DOM initializer, preserving early storage capture without the observed null-parent startup exception.

Validation:

• 2,143 Node/DOM tests, 132 chart-storage browser scenarios, test lint, build, syntax and diff checks passed; final read-only review passed.
• Real IndexedDB scenarios cover value semantics, admission, abort, native continuation, concurrent tabs, clear, stop and reload.
• The complete artifact passed an anonymous Chrome smoke: chart ready, zero page errors, five optimized commits, six writes and thirteen skipped writes. After stop, four native commits completed with frozen optimizer counters.
• The existing local Tampermonkey orderbook entry was updated and read back exactly: SHA-256 f281dd8c7b5f384e6f05ef1f7e837201923a0194d7fcdebd135997ee8e766f64. The temporary independent entry was removed and its absence verified. The integrated page passed exact loaded-source and MAIN_WORLD checks, rendered chart/panel inspection, 37 committed batches with 10,970 skipped writes and zero failed/aborted/pending batches after stop. A separate hard-refresh capture passed the canonical Stage 3 verifier: the script parsed before DOMContentLoaded, the complete loaded source matched once, and multiplier 1 → 2 → 1 emitted no financial requests. All test tabs were closed.

This is page-scoped mirror-write optimization, not a Chrome-wide patch or a claim that all historic blank pages had one cause. Batch atomicity differs from native per-key writes; cross-database atomicity, newest-revision ordering and long-running stability are not promised. See docs/binance-chart-storage-development.md for the exact boundary and docs/binance-futures-blank-page-diagnosis.md for measured evidence.